Hello all! I work with geospatial data infrastructure at scale, so I end up thinking a lot about how n-dimensional datasets are stored. I've been spending a lot of time writing about raster array storage and data formats with a colleague, and one hard concept to talk about and for people to understand is data linearization. It's especially hard to then communicate how different chunking and linearization strategies affect the efficiency of read access for different access patterns.
On a whim, as a backend engineer, I asked claude if a static webpage could be used to visualize this in an interactive manner, and what it spit out was pretty amazing. A few days and much iteration later, and I think what has resulted is a rather useful tool for understanding chunking, linearization, and the impacts of each on data access. So I wanted to post it, let people play with it, and see if others find it interesting and/or useful too!
On a whim, as a backend engineer, I asked claude if a static webpage could be used to visualize this in an interactive manner, and what it spit out was pretty amazing. A few days and much iteration later, and I think what has resulted is a rather useful tool for understanding chunking, linearization, and the impacts of each on data access. So I wanted to post it, let people play with it, and see if others find it interesting and/or useful too!